Lessons Learned from Industrial Chemical Accidents: A Case Study of Indian

Oil Corporation (IOC) Terminal, Jaipur, India

Ravi K. Sharma1*, Bhola R. Gurjar2, Henry Keen Peng. Ho3, Santosh P. Ghuge4, Rajat Agrawal5
1*
Centre for Excellence in Disaster Mitigation and Management, Indian Institute of Technology
Roorkee, Roorkee 247667, India
2
Department of Civil Engineering, Indian Institute of Technology Roorkee, Roorkee 247667,
India
3
IGnesis Consultants Pte Ltd, 408868, Singapore,
4
National Environmental Engineering Research Institute, Nehru Marg, Nagpur 440020, India
5
Department of Management Studies, Indian Institute of Technology Roorkee, Roorkee 247667,
India

Ϯ Presenter E-mail: ravis.mdt2015@iitr.ac.in

Abstract

Fire and explosion hazards in industrial storage units have gained a considerable attention in
recent years. Indian Oil Corporation (IOC) storage terminal accident in Jaipur, India, is a recent
example of Vapor Cloud Explosion (VCE) and fire accidents preceded by Buncefield (2005) and
Puerto-Rico (2009). On 29th October 2009, a leak of gasoline occurred in the IOC storage
terminal. Long delay of 80 minutes in ignition led to a huge vapor cloud covering an area of
180,000 m2 over the entire installation and subsequent ignition triggered strong VCE with
intensity more than 200 kPa. Eleven people lost their lives, more than 150 people were injured
and a property loss of approximately U.S. $60 million was reported. The Individual and Societal
Risk has been quantified and found that risk does not lie in the unacceptable region, but in the As
Low As Reasonably Practicable (ALARP) region where substantial measures for a risk reduction
were needed.

The incident has left many safety issues behind which must be repeatedly addressed. It reveals
that adequate safety measures were either underestimated or not accounted seriously. This article
highlights the aftermaths of IOC incident and addresses challenges put forward by it.
Furthermore, a comparative study is performed between such incidents to analyze the similarities
and how they could have been avoided. Therefore, electronic-Incident Command System (e-ICS)
based emergency response planning is an integral and essential part of the safety and loss
prevention strategy and comprises of the actions taken to manage, control and mitigate the
immediate effects of an incident.

Introduction
Despite of up to date safety know-hows, industries all over the world still confront frequent
explosion and fire hazards [1–3]. The accidents may occur due to different reasons extending
from malfunctioning of an installed mechanical device to mistake committed by any personnel.
The subsequent consequence of such failure results in abrupt release of stored fuels forming
vapor cloud around a facility. Depending on flammability and the availability of an ignition
source, such a vapor cloud may lead to a massive explosion which has been referred as Vapor
Cloud Explosion (VCE) in the process safety literature [1–5]. In the last decade many of such
major accidents in storage sites are reported [6]. There have been two large-scale incidents viz,
Puerto Rico, (USA) in 2009 and Buncefield (UK) in 2005 which have exhibited striking
similarities with Indian Oil Corporation Limited (IOCL), Jaipur, (India) storage terminal
accident [7-9]. A wide range of similarities in VCE and subsequent fire have been observed
among these accidents. Only the Buncefield and IOCL, Jaipur investigation reports have been
published till date [1, 3]. One expects that the overpressure generation as a result of such VCE
often lie in the range between 20 and 50 kPa [4,5]. Such overpressures are possible to evaluate
by the existing models in the open literature. But, the peak side on overpressure estimated after
the IOCL, Jaipur accident [9] of >200 kPa was not expected based on the existing knowledge on
VCE. There are lists of parameters required to be considered behind this excessive overpressure
generation. Some of them like congestion in the form of trees and bushes can have a significant
contribution on the Deflagration to Detonation Transition (DDT) [10-14]. The intensity of
disaster due to explosion becomes severer when DDT is seen as a result. Hence, a careful review
of the safety measures related to conditions favoring the generation of excessive overpressure are
necessary to be executed.

In the second part of this paper, safety distance estimation from after explosion scenario
which is very often a large pool fire is presented. Strength of explosion, number of tanks and
surrounding conditions could lead to single or multiple fire. Subsequent explosions also lead to
the engulfment of surrounding tanks / containers and contribute towards a violent fire. Whenever
such a scenario has occurred the seen consequences were even worse. The affected area can be
of the order of a few meters to several kilometers depending on the magnitude of the incident.
Large-scale accidents have a potential to harm the on-site and off-site population. In the third
part, paper presents an assessment of Individual Risk and Societal Risk (IRSR) associated with
the cumulative effects of explosions and fires. However, the total risk at the IOCL, Jaipur
accident does not lie in the unacceptable region, but in the As Low As Reasonably Practicable
Region (ALARP) where substantial measures for a risk reduction were needed. The
consequences in and around the terminal were high, which might be due to the failure or absence
of certain precautionary measures. Moreover, according to Buncefield Major Incident
Investigation Board (BMIIB) [1], land use planning (LUP) is responsible to the risk on the site.
BMIIB suggested that LUP should be based on the risk level and more attention should be paid
to minimize the risk to the surrounding population. Therefore, a detailed assessment and analysis
of risk is required that can help determine adequate safety measures to avoid such fatal
incidences or reduce their severe effects.
Successively, Immediate and effective response to an accident site is, of course, necessary to
reduce the severity of accidents, loss of life and the possibility of the loss of the future
productivity of the storage site [15]. Thus, the main focus in the management of emergencies has
been on resources and logistics; in other words, having who and what you need, when and where
you need it to encounter the crisis within an urgent time frame.
Brief summary of the major accidents
An introduction of major accidents occurred during the last decade are presented in this section.
There might be other causalities but we limit our discussion only up to: a. Buncefield (2005), b.
Puerto Rico (2009) and c. IOCL, Jaipur (2009).

Buncefield (2005, UK)

On the early morning of 11 December 2005 one tank (912) at Buncefield storage depot started
receiving excess amount (more than its capacity) of unleaded petrol leading to an overflow of the
tank and subsequent collection of petrol in the bund [1]. The scenario is shown in Fig. 1. The
failure of automatic safety switch alarm allowed the tank 912 to fill at a rate of more than twice
of that in normal operation [1]. By calculating the overpressure of liquid in tank 912 from the
following relation between pressure drop and flow rate of an incompressible fluid, △p ∝ V2, it
reveals that the tank was overflowing at a liquid overpressure of 2.2 times of the pressure under
normal operating conditions. It is likely that this liquid overpressure certainly contributes
towards magnitude of formation of vapor cloud under stable wind conditions [1] and subsequent
explosion (gases) overpressure. The existing understanding on the overpressure of a vapor cloud
explosion rather underestimates the reported intensities of overpressures at Buncefield. One
would have estimated an average value of overpressure of 5 kPa [4,5] for such vapor cloud
explosion which simply does not represent the devastating recorded (determined from damage
analysis) overpressure of >200 kPa [4,5]. The diameter of the vapor cloud was estimated to be
about 391 m and the high speed rotating machines in the pump house are postulated to be the
potential source (initial) of ignition [1,4,5]. The effect of overpressure could be felt as far as 2
km from the center of explosion [1]. Fortunately, no fatalities took place though light and serious
injuries could not have been avoided.

Fig. 1. Buncefield accident (2005, UK) [1]

Puerto Rico (2009, USA)
A similar scenario to Buncefield accident was seen on 23 October 2009 at Caribbean
Petroleum Corp., Bayamon, USA, where also an uncontrolled flow of petroleum products mainly
gasoline led to the formation of a vapor cloud and finally resulted in heavy explosion
overpressures (Richter scale 2.8) [2]. The scenario is shown in Fig. 2. The investigation report
currently being exercised by the Chemical Safety Board of the USA and thus many details on
this accident are not available to date [2]. The feedback of eyewitnesses/ evidences also supports
the similarity to the Buncefield disaster [2].

Fig. 2. Puerto Rico, USA (2009) storage fire accident [2].

IOCL, Jaipur (2009, India)

On 29 October 2009, a petrol terminal in the Indian Oil Corporation (IOC) in Sitapura
industrial state (near Jaipur) caught fire following an explosion and continued to rage until a
week [3] (Fig. 3). The oil industry safety directorate of India has investigated the incident [3].
They found potential similarities with Buncefield except the ignorance/ mistakes done by
personnel which were additional reasons reported behind the IOC, Jaipur accident. However,
from a scientific point of view the occurrence was similar to the above mentioned two incidents.
The overflowing of a tank, formation of vapor cloud in still air (low wind speeds), ignition
triggered by the pump/generator station and finally the spreading of fire over the other tanks [3].
 Fig. 3. Sitapura, India (2009) storage fire accident [3].

Accident and Its Aftermath

The IOC, Jaipur, India, incident was one of the major petroleum storage terminal
accidents in India. During transfer operations through a pipeline to another terminal, a series of
vapor cloud explosions (VCEs) had occurred as a result of the uncontrolled release of gasoline
from the so-called hammer blind valve of Tank 401-A over a period of 80 min before ignition of
the resulting flammable mixture. The total amount of gasoline released was 2,000 metric ton,
which resulted in a formation of 81 metric ton of vapor cloud covering an area of 180,000 m2 [9].
Subsequently, the ignition of a flammable mixture had resulted in massive explosions and
intense fires. A series of powerful explosions was heard up to 32-km away from the terminal.
Seismological measurements reported that one of the VCE was equivalent to an earthquake with
the intensity of around 2.3 on the Richter scale [3].
Due to such massive explosions, the entire installation was destroyed and the buildings in
the immediate vicinity were heavily damaged. The associated blast wave caused windowpane
breakages, which were found up to 2 km from the terminal [3]. After one of the major explosions
there was a fire that engulfed 11 large storage tanks. The fire burned for a week, destroying most
of the site. The vegetation around the storage facility was completely consumed by the fire. The
management of IOC had taken a considered decision to allow the petroleum products to burn out
to avoid further possibilities of accident in the installation thus ensuring safety of the public.

Casualties
Eleven people lost their lives in the accident (six from IOCL and five outsiders) and more
than 150 peoples were injured. In addition to this, about 5,000 people in the nearby surrounding
area had to be evacuated from their homes [3]. This makes it one of the most fatal accidents that
have occurred during the last decade in the petroleum industry.
Hazard criteria

In general, a storage site for flammable liquids is designed as per the standard norms
prescribed by the regulating bodies. However, some industries have their own hazard mitigation
plans. In either case, consequences arise from explosion and fire must be addressed. Most
importantly, the damage to people and vicinity which may be caused by the explosion
overpressure [4,5] or/and by thermal radiation [16-17] from fires are required to be specified
before installation and operation. Technically, explosion refers to vapor cloud explosion or gas
explosion and fire refer to single/multiple large pool fires in this article. This will be discussed as
follows.

Evaluation and analysis of an accident

In the present analysis, the first event (i.e. the accidental release) has been analysed and
then the modelling up to the VCE, the final event, has been carried out to identify all the possible
consequences of explosion.
The release rate

As reported by the Independent Inquiry Committee [3], the initiating event of the
sequence was caused by the accidental releases of gasoline from the 0.25 m diameter outflow
pipe. This resulted in the leakage of substantial amount of flammable liquid. In the first case, the
mass discharge rate from the Hammer Blind Valve was determined by the Equation (1) [18].
This assumed the fraction is represented by a discharge coefficient, C D , and accounts for the
pressure due to the liquid head above the hole, h L .

g𝑃
𝑚 = 𝜌𝜗𝐴 = ρ𝐴CD �2 � 𝑔�𝜌 + gℎ𝐿 � (1)

where 𝑚 is the mass discharge rate, 𝜗 the fluid velocity, 𝐴 is the area of the hole, 𝜌 is the
density of the liquid at 30°C and 𝑃𝑔 the gauge pressure at the top of the tank (for a tank open to
the atmosphere 𝑃𝑔 = 0 ), and g is the gravitational constant. The density of gasoline is 740 kg/m3
at NTP and the liquid head above the hole is 14 m. By using Eq. (1) the estimated mass
discharge rate or release rate is 323 kg/s.

For liquids that are accelerated during the release, such as in a jet, a common approach is
to assume an isentropic path. If the liquid temperature is less than the normal boiling point, the
flash fraction is zero [19].
 The Evaporation Rate

The calculation of the vapour mass involved in the explosion is crucial for assessing the
consequences of such accidents. A more accurate result can be obtained by computing the
evaporation rate of gasoline during the accidental release.
The evaporation rate per unit area m mass can be calculated by considering two contributions
that are derived, respectively from pool area and from the liquid falling into the pool. The
vaporization rate for this situation is not as high as for flashing liquid or boiling pools, but can be
significant if the pool area is large. Atypical approach is to assume a vaporization rate of the
form [20 ]:

𝑀𝑊𝑘𝘨 𝐴𝑃 𝑃 𝑠𝑎𝑡
𝑚𝑚𝑎𝑠𝑠 = (2)
𝑅𝗀 𝑇𝐿

where MW is the molecular weight of the evaporating material, k g is the transfer

coefficient, A P is the area of the pool, P sat is the saturation vapour pressure of the liquid, R g is
the ideal gas constant, and T L is the temperature of the liquid. As far as the pool evaporation term
is concerned, the mass transfer coefficient can be computed as:
As far as the pool evaporation term is concerned, the mass transfer coefficient can be computed
as:

𝑘𝘨 = 0.002 × 𝑣 0.78 𝐿−0.11

𝑃 = 1.7× 10-3 (3)
where, 𝜗 is the wind velocity at a height of 10 m from the ground and
𝐿𝑃 is the pool length. The gasoline evaporation rate has been estimated as 17 kg/s by using Eq. 2.
It is estimated that in 80 minutes of uncontrolled release resulted in about 81 tonnes of gasoline,
which might have formed an adequate vapour cloud of gasoline for a massive explosion.

The Formation and Size of the Flammable Cloud

The formation of large vapour clouds may be due to the following factors, which increase the
cloud size for dense and neutrally buoyant vapour clouds [21].

• Still wind conditions.

• Delayed ignition potential increase due to better control over ignition sources.
• Ease/speed of detecting loss of containment.
• Significant delay in arresting the release (that increases the radius of the impact zone).
 The nearest meteorological measurements indicate that on evening of the day of the
accident, the weather was calm and stable, with an air temperature close to 30°C and 25%
humidity. The weather on the evening of the release was Pasquill stability category D, with a 1.5
m/s wind speed [3].

The amount of air entrained into the cascade by momentum exchange is sensitive to the
liquid mass density, which, in turn, requires knowledge of the width of the spray zone [22]. In
the case of the IOC Jaipur incident, the vapour cloud formation was favoured due to air
entrainment, dispersion from the falling strings of gasoline and evaporation of the gasoline in the
bund. Additionally, the topography of the surrounding land and the blocking due to undergrowth,
storage tanks and the plant affected the spreading of vapour cloud.

However, the vapours would have spread as a gravity current, mixing with air at the
leading edge and top surface of the cloud, whereas the lower part would have remained stratified
and fuel rich. It seems reasonable to suggest that the centre of the cloud would be deeper and
richer in fuel than the edges [22]. Once initiated, the flame would flash through the flammable
regions, leaving the rich mixture to burn more slowly as diffusion flames. The flammable limit
corresponding closely to the top of the mist layer may not hold and needs to be justified by the
thermodynamics of the local cloud composition and atmospheric humidity[22]. Complete
appreciation of the mechanism of this cloud formation has proved difficult to achieve. Thus, the
source term for the vapour dispersion contains many uncertainties and inherent difficulties.

The estimation using Phast Risk 6.7 [23] software shows that the total area of the cloud
was of the order of 180, 000 m2 and extended to a distance of almost 500 m, with an estimated
height of 2 m over most of the area [9]. The wind direction at the time of incident was 340°
(NNW direction), with a stability class of D. Fig. 4 shows the vapour cloud dispersion with a
varying concentration of material, towards the south east (SE) direction using DNV Norway
based Phast Risk 6.7 software [23]. It was a massive cloud, both absolutely and compared with
the clouds observed in other incidents, and the cloud size is an important reason why the
explosion was so large. With regard to the development of the vapour cloud, the height of the
cloud is also important. This study was undertaken to understand how the vapour cloud spread
over such a large area and to provide data that could be used for explosion modeling studies.
 Fig. 4 Dispersion of vapour cloud in SE direction by using Phast Risk 6.7

Vapor Cloud Explosion (VCE)

A consequence in form of a VCE can be seen as a combination of many factors. The

overpressures generated due to VCE (with regard to the incidents considered here) are primarily
influenced by the following parameters:

1. Flammability and quantity of fuel

2. Degree of confinement/congestion
3. Source and strength of ignition
4. Weather conditions.
Most of the empirical models developed to date are typically based on the above four
parameters. Absences of a single factor alter the probabilities and extent of occurrence to a great
extent. For example, if a vapor cloud is formed at a location but could not find a source of
ignition so it may not be harmful or a situation where the wind speed is high and the vapor are
carried away thus minimizing the risk of accumulation [4–6, 10]. Many of such possibilities exist
and a careful review of permutation and combination is required for the risk assessment studies.
There has been significant study done in the past and recently to establish common conclusions
on the resulting overpressure from IOC, Jaipur VCE [9]. While many of them report that the
excessive overpressure was not usual for the common understanding on VCE till date a number
of papers also indicate the likeliness of such high overpressure generation [5,6]. Regardless of
which is true we try to examine the possible applicability of available models today. There are
three different methods considered in the relevant literature to estimate the overpressure for a
VCE. Each of them is briefly described below.

TNT equivalent method

A common approach for determining the damage caused by a given explosion consists in
estimating the “TNT equivalency”, i.e. the mass of TNT that would produce the same degree of
damage [24]. The main features of TNT and other high explosives have been extensively studied
and are therefore reliable references. Since significant experimental data on the explosion
characteristics of TNT are available it is easier to extrapolate the scenario with TNT. The
following Eqs. (4) and (5) describe the same.

𝜂𝑀𝐸𝑐
𝑊= (4)
𝐸𝑇𝑁𝑇

𝑅
𝑍∗ = 1/3 (5)
𝑊𝑇𝑁𝑇

Fig. 5. Overpressure produced by a VCE vs. scaled distance [24].

 where W is the equivalent mass of TNT, ƞ is an empirical explosion efficiency, M is the
mass of hydrocarbon, 𝐸𝑐 is the heat of combustion of flammable gas, and 𝐸𝑇𝑁𝑇 is the heat of
combustion of TNT. Eq. (4) has a yield factor ƞ which depends on the degree of congestion on
the site. Typical values based on experience for ƞ are between 3% and 5%. However, for
stoichiometric proportions and heavily congested area a value of 20% is specified [14, 25-27]. In
Eq. (5), R (m) is the distance from the centre of explosion. The estimation of on side
overpressure can be made with the help of plot for pressure on site vs. scaled distance 𝑍 ∗ (m
kg−1/3) as shown in Fig. 5. For the sake of initial estimation this method could be utilized. For
example when this method is applied to IOC Jaipur it estimates an overpressure of ~9 kPa at a
distance of 2 km from the source of explosion provided the yield factor is 20%. Even when the
lowest value of yield factor is considered, i.e., 3% .
The biggest drawback of this method is that it is not applicable to overpressures of >100
kPa and small distances 𝑍 ∗ as the local pressure developed in case of TNT is much higher than
that in case of flammable gas explosion waves which travels to a larger distance. Therefore, for
far field regions it does not estimate the overpressure correctly. When this method is applied to
IOCL Jaipur accidents it estimates the overpressures (at 300 m) to be 13 kPa, respectively,
depending on the mass of flammable fuel responsible for explosion.

Baker and Strehlow method

This method is based on the Mach number MW (flame velocity), reactivity of fuel and
level of congestion and confinement [5, 28]. The Eqs. (6), (7) and (8) for maximum overpressure
𝑃𝑚𝑎𝑥 , dimensionless average side on pressure 𝑃�𝑠 and the scaled distance 𝑅� are

𝑊𝑀2
𝑃𝑚𝑎𝑥 = 24 (1+𝑀 (6)
𝑤)

𝑃
𝑃�𝑆 = 𝑚𝑎𝑥 (7)
𝑃
𝑎

𝑃 1/3
𝑅� = 𝑅 �𝐸𝑎 � (8)
1

When a medium reactivity, a Mach number M W of 0.55, a total available energy E 1 (J)
and a high level of congestion are assumed an overpressure of 50 kPa results. The 𝑃�𝑠 is plotted
against 𝑅� in Fig. 6. The same overpressure values are also estimated for the other two accidents
when a Mach number of 0.55 is assumed (generally valid for hydrocarbons).
 Fig. 6 Overpressure produced by a VCE vs. scaled distance [24]

TNO (multi energy method)

This is the most widely used method in Europe [5, 29]. The challenge in this method is to
determine an appropriate explosion source strength. The Eq. (9) describes the maximum
overpressure as follows:

𝐿 2.75
𝑃𝑚𝑎𝑥 = 0.84 �𝑉𝐵𝑅 𝐷𝑓 � 𝑆𝐿2.7 𝐷0.7 /84 (9)

Where 𝑃𝑚𝑎𝑥 : maximum overpressure in kPa; VBR: Volume Blockage Ratio (%); 𝐿𝑓 :
flame path length (m); D: average obstacle diameter (m); 𝑆𝐿 : laminar burning velocity of
flammable mixture (m/s).The dependence of dimensionless overpressure Eq. (7) on distance Eq.
(8) is shown in Fig. 7. When applied this method to IOCL, Jaipur with VBR 4%, 𝐿𝑓 =50 m, D
=0.3 m and S L =0.46 m/s for hexane [3] it estimates the overpressure to be >2000 kPa.
 Fig. 7. Overpressure produced by a VCE vs. scaled distance [29].

Estimation of maximum peak overpressure

In the present study, the estimation of the maximal peak overpressure ∆P max by an
unconfined vapour cloud explosion (UVCE) and by a (partially confined) VCE was mainly
focused. Hailwood et al., [30] reported that the course of a UVCE should be treated as a
deflagration (∆P max < 1 bar) or as a detonation (∆P max > 1 bar). When taking this into account, a
formula can be derived for a spherical pressure waves for an unconfined and partially confined
vapour cloud explosion between the flame front velocity 𝑢𝐹 and the maximum peak over-
pressure ∆P max is determined from Eq. 10.

2 𝑢 2
𝛥𝑃𝑚𝑎𝑥 2𝛼�1−1�ε1 � � 𝐹�𝑐𝑠 �
�𝑃 = 𝑢 (10)
𝘢 1+�1−1�ε1 � 𝐹�𝑐𝑠

Where, the expansion ratio ε1 (measure of the energy release rate) is determined from Eq. 11:

𝜌0 𝑉
ε1 ≡ �𝜌1≡ 1�𝑉 (11)
0

The range of ε1 ≈ 7 to 8 [31] for stoichiometric hydrocarbon-air mixtures.

 The other important parameters of the VCE are the shock wave velocity, maximum
dynamic pressure and the maximum reflected overpressure [32]. The shock wave velocity in air,
U, is calculated by using Eq. 12.

1/2
0
𝑈 = 𝐶0 + �1 + 6𝑃 �7𝑃� (12)

where, 𝐶0 is the speed of sound in air, and P and 𝑃0 are atmospheric pressure and
maximum overpressure, respectively. The estimated shock waves velocity which is generated
due to massive explosion travel with a speed of 488 m/s [9].

Dynamic pressure 𝑞 0 refers to the transformation of kinetic energy of the wind generated
due to explosion into pressure energy when encountering a solid surface in its path. For
explosion in air, the maximum dynamic pressure𝑞 0 can be expressed as Eq. (13) [32].

0
5 (𝑃0 )2
𝑞 = (13)
2 7𝑃 + 𝑃0

where 𝑞 0 is the maximum dynamic pressure. Lastly, it is important to consider the

maximum overpressure due to wave reflection. When the pressure wave hits a solid surface not
parallel to the propagation direction, a reflection is produced and the reflected pressure varies not
only with the value of 𝑃0 , but also with the angle of incidence. The maximum overpressure
takes place when the pressure wave hits the overpressure experienced and that which takes
places on a surface parallel to the direction of propagation.

 7P + 4P o 
( )
P r =2 P 
o o
 (14)
 7P + P
o


where, (Po) r is the overpressure produced on a surface perpendicular to the direction of

propagation as a consequence of the reflection and r denotes the ‘reflected’ overpressure. Eq. 14
shows that the maximum reflected overpressure is at least double 𝑃0 and could become 8 times
greater. However, for weak explosion 𝑃0 can be smaller as compared to atmospheric pressure.
The calculated dynamic pressure and reflected over pressure with respect to maximum
overpressure (˃1 bar) are 0.32 bar and 2.7 bar, respectively [9]. The damage caused by this
explosion resulted in further loss of containment and the subsequent fires involved a number of
fuel storage tanks on the site.

Deflagration-to-Detonation Transition (DDT)

 Unlike deflagration, detonation is self-sustaining and propagates across the open areas if
the vapour cloud concentration is within the detonable limits (which are generally similar to the
flammable limits for common hydrocarbons). As a result, the directional indicators would be
more widespread [33]. A critical condition for a transition from deflagration to detonation (DDT)
in a duct caused by attaining a maximum turbulent burning velocity should be high enough for
the gas velocity ahead of the flame to generate a shock wave sufficiently strong [34].

In the IOCL Jaipur site, the most probable cause of the detonation was a flame entering
either the pipeline area control room or the pipeline pump house located at north east corner of
the site, causing a confined or partially confined explosion that might have initiated a detonation
as it vented from the building [33]. The damage of the pipeline control room building and
Pipeline pump house are shown in Figs. 8 and 9. As shown in Fig. 8, damage to the south side of
the building was much more severe than on the north side of the building, where there was a
complete collapse of the building, indicating the propagation of waves towards the pipeline
division from the south side. Fig. 8 shows the damage to the pump house from the south side.
Trees bent towards the northeast direction, as examples of directional indicators, are also shown
in Fig. 10.

(a) (b)
Fig. 8 Control room in the Pipeline Division area (a. north side b. south side) [33]
 Fig. 9 Damaged Pipeline pump house from the south of the building [33]
There are two probable descriptions that can validate the pipeline control room damage.

• There is a clear dividing line between the high pressure damage to the south side and the lower
level of damage on the north side. This finding is also supported by the apparent lack of damage
to the trees on the north side of the control room that can be seen in Fig. 8a.

• It is notable that the collapse of the roof downwards on the south side (Fig. 8b) does not appear
to be consistent with an internal explosion that vented outwards from the north side building. The
flames venting from the building might have resulted in a transition to detonation and the high
external pressure could have pushed the partially failed roof downwards. This description could
be considered to be physically plausible.

The directional indicators point to the source of the detonation being located in the
Pipeline Division area in the northeast corner of the site, as illustrated in Fig. 11. The arrows
indicate the approximate directions indicated for each area of the site [33].
Fig. 10 Directional indicators: bent tree (towards the northeast direction) towards the pipeline division

Fig. 11 Overview of the directional indicators and estimated cloud boundary (yellow line)
[33]

There was a confined explosion in the control room that could have eventually led to a
transition to detonation in the vapour cloud on the south side, or it could have enhanced flame
propagation towards the pipeline pump house further to the south, with a detonation being
initiated by an explosion in this building. This hypothesis can be supported because at the
downwind side, the wind flow reattaches to the ground and the mean velocity remains lower than
it is on the upwind at the same height above the ground. All along this wake, turbulence is higher
 than the upwind side values. Thus, flammable material near a building can have higher
concentrations in the building wake than in the absence of the building. Enhanced turbulence
accelerates flames near and far downwind of the building. The directional indicators would then
be produced by a combination of asymmetric propagation of the detonation combined with direct
overpressure effects.

A deflagration to detonation transition due to trees along the north wall of the pipeline
division has not been considered because there were no dense bushes at a lower level and some
gaps were found in the tree line. These gaps in the tree line might have decelerated the transition.
Whereas, in the case of the Buncefield analysis, the possibility of the detonation occurring as a
result of flame acceleration in trees does not appear to be consistent with the evidence, and it was
found that directional indicators could be explained by a detonation propagating through the low
lying vapour cloud [11]. The evidence obtained from the IOCL Jaipur site has a high degree of
consistency with the observations made following the Buncefield incident, both in terms of
overpressure damage and directional indicators. Table 1 lists some of the important details
concerning the accidents.

Table.1 A summary of the facts of major incidents that have similarities with the Jaipur IOCL gasoline
release accident
Explosion Overpressure
Storage Capacity (m )
Location of Accident

Intensity on Richter
Delayed in ignition

Loss
Quantity Released

Fire Lasted (days)

Cloud Cover (m )
2

destroyed / total
Damages (tanks

Economic Loss
& Date

Cause

(min.)

(tons)

Scale
(bar)

Injured

tanks)
Death

($)
Buncefi
eld 273×1 Overfilli 4 12×1 ≈ ≈ ≈ N 1.5×1
UK, 3 300 4 43 22/41 9
0 ng 0 0 1.3 2.4 4.5 il 0
Dec.
2005
IOC
Jaipur, 110×1 8 200 18×1 ≈ ≈ 15 6
India, 3 Leakage 4 >2 11 11/11 6
Oct. 0 0 0 0 2.3 11 0 ×10
2009

Fire hazard criteria

Hazard from a fire is generally defined in terms of emitted thermal radiation to the people
and objects [16]. The fires in the considered accidents burn in form of large pool fires. The
diameter of a pool can range in the order of some meters. In the open literature a significant
 amount of work has been reported on thermal radiation characteristics of different hydrocarbon
pool fires [16-17, 28, 35]. The hydrocarbon in the considered accidents was primarily gasoline.
The amount of thermal radiation emitted by a large gasoline pool fire (D = 24 m) at different
distances is shown in Fig. 12.With the assumption that thermal radiation from the surface of a
turbulent fire (D≫1 m) is saturated one can estimate the safety distance. There are several
methods described in the literature [1]. We will also use them for the accidents described above.

20
18 Gasoline (D = 24m);
16 (H/D)max = 1.5
Raditation Level (kW/m²)

14
12
10 Point Source…
PHAST 6.51
8
6
4
2
0
0 50 100 150 200 250 300
Distance Downwind (m)

Fig. 12 Thermal radiation variation with relative from a fire

Flame characteristics and modelling of pool fires

The pool fire characteristics, like the following description of the consequence model
given below, have been used to estimate the flame height, flame temperature and thermal
radiation.

Flame Geometry

The geometry of a flame depends mostly on flame pool diameter, flame length, mass burning
rate, temperature and the flame radiative properties. These properties are characteristically taken
as averaged in time. The measurements derived from different assessments for the influence
factors and the geometry of large flames is shown below

Relative Flame Height

The flame height is generally taken as the maximum visible height or the time-averaged
visible height [36]. The time-averaged relative (𝐻 � ⁄𝐷) and maximum relative ((𝐻
� ⁄𝐷)max)
visible flame height are dependent on the Froude number (𝐹𝑟𝑓 ) and the dimensional wind
∗
velocity (𝑢�𝑊 ) and can be estimated by the following correlations [30]:
 ��
𝐻 𝑏 ∗𝑐
𝐷 = 𝑎 𝐹𝑟𝑓 𝑢�𝑊 (15)

and

�𝐻�𝐷�
𝑐
= 𝑎 𝐹𝑟𝑓𝑏 𝑢�𝑊
∗
(16)
𝑚𝑎𝑥

There are more correlations with many empirical parameters, including a, b, and c, which
are experimental parameters and are given in Table. 2 [24].

Table. 2. Parameters for the determination of the dimensionless visible flame heights used in Eq. (15,
16) [37]

Correlation a b C Comment

Munoz 1 8.44 0.298 -0.126 Measured on gasoline and diesel pool fires:
(H⁄D) max
R

Munoz 2 7.74 0.375 -0.096 � ⁄𝐷)

Measured on gasoline and diesel pool fires: (𝐻

The height of the visible flame is a function of the pool diameter and the burning
velocity. For the IOCL Jaipur incident, an assessment of the maximum, visible and relative flame
heights of gasoline tank fires was conducted assuming that the‘c’ parameter in Eq. (17) was zero
because there was no wind effect. The modified equation can therefore be written as:

𝑏
����
𝑚 ͌𝑓
(𝐻 ⁄𝐷)𝑚𝑎𝑥 ≈ 𝘢Fr𝑓𝑏 = a �𝜌 � (17)
𝑎�𝘨𝐷

Thus, the estimated (𝐻 ⁄𝐷)max ratio for the gasoline tank (D = 24m) fire is 1.5. For a
large hydrocarbon pool fire where D ≥ 9 m, the time-averaged relative flame height (𝐻 � ⁄𝐷) is
calculated using Eq. (18) [30] and Table. 1 and approximates to
0.375
����
𝑚 ͌𝑓
� ⁄𝐷)calc ≈
(𝐻 𝘢Fr𝑓𝑏 = 7.74�𝜌 � = 0.9 (18)
𝑎 �𝘨 𝐷

With ����
m′′f,max (D=24 m) ≈ 0.055 kg/ (m2s) for a gasoline pool fire, 𝜌𝑎 = 1.29 kg/m3, and
the parameters a and b from Table 1, the calculation using Eq. (17, 18) results in

0.9 ≤ (H/D) max,calc ≤ 1.5 (19)

An empirical relationship was observed between the maximum and average flame height.
Thus, a single correlation could be used to estimate both dimensions [38]:

� ⁄𝐷
(𝐻 ⁄𝐷)max ≈ 1.6 𝐻 (20)
The empirical relationship in Eq. (20) was also considered valid for the IOCL Jaipur tank fires.

Flame Temperature

The flame temperature is a function of time and height, as described by Planas and Casal
[39]. The correlation used for the flame temperature is given by the following equation (Eq. 21):

104 .𝑡
T𝑓 (𝑡, ℎ) = (34+210 ×H+8.51×t) + 298 (21)

In the IOCL Jaipur incident, the estimated flame temperature of the gasoline tank (D =
24m) was approximately 1230K, which lies within the range (1100K-1240K)reported by various
researchers for large-scale gasoline pool fires [17, 40 - 43] .

Individual Risk and Societal Risk Assessment

If the accident is severe, it can cause serious injuries or fatalities to surrounding people.
Therefore, a detailed assessment and analysis of risk is required that can help determine adequate
safety measures to avoid such fatal incidences or reduce their severe effects. According to
BMIIB [1], land-use planning (LUP) is responsible for the level of risk on the site. BMIIB
(2008) also suggested that LUP should be based according to the risk level and more attention
should be paid to minimize the risk to the surrounding population. CCPS [19] gives guidelines to
estimate the individual and societal risk associated with different incident outcome cases from
major chemical industrial accident.

In this paper, an attempt has been made to make reasonable assumptions to provide a
more realistic estimate and analysis individual and societal risks due to an explosion and/or fire
at petroleum oil storage terminal. The release modeling has been carried out by classifying
affected areas in and around the terminal, incorporating population density, and by focusing on
the time sequence of events.

Method of Analysis and Input Data

There are two kinds of risks to people, i.e., individual and societal risk. Evaluation of
individual and societal risk is the key point for the probabilistic safety assessment of the storage
terminal. The individual risk is defined as the probability of death per year of exposure to an
individual at a certain distance from the hazard source. It is usually expressed in the form of iso-
risk contours around the source of hazard [44]. Whereas, societal risk as “the relationship
between frequency and the number of people suffering from a specified level of harm in a given
population from the realisation of specified hazards” [45]. Societal risk is presented in terms of
F/N curves, where N is the number of fatalities and F is the frequency of N or more fatalities.
Many countries such as Australia, the Netherlands, Malaysia and UK employ numerical criteria
in determining acceptability of risk in terms of safety zones. The ALARP principle is developed
by the Health and safety Executive of the UK [46]. It states that risk should be reduced to “As
low as reasonable practicable” (ALARP) level. The ALARP principle divides risk into three
bands: intolerable risk at the higher end, negligible risk at the low end, and the tolerable risk in
between. As shown in Fig. 13. Maximum tolerable individual risk for workers is 10-3 per year
whereas for members of public it is 10-4 per year. Risk in the middle region can be tolerated as
long as all cost- effective measures to reduce risk have been put into place. The cost in reducing
risk should not exceed the benefits gained in reducing risks. A process with risk in the tolerable
risk region must demonstrate that the lowest risk has been achieved by taking into consideration
cost versus risk reduction criteria.

Fig. 13. The ALARP principle, developed in the UK (HSE, 2001)

Individual Risk at a geographical location x, y is given by AIChE/ CCPS [19] as follows

(Eq. 22):
𝑛
𝐼𝑅𝑥,𝑦 = �𝑖=1 𝐼𝑅𝑥,𝑦,𝑖 (22)

where, 𝐼𝑅𝑥,𝑦 is the total individual risk of fatality at geographic location x, y.

𝐼𝑅𝑥,𝑦,𝑖 is the individual risk of fatality at geographical location x, y from the incident
outcome case i (chance of fatality per year), n is the total number of incident outcome in the
analysis from the industrial area. 𝐼𝑅𝑥,𝑦,𝑖 can be estimated using the following equation (Eq. 23).

𝐼𝑅𝑥,𝑦,𝑖 = 𝑓𝑖 × 𝑝𝑓,𝑖 (23)

Where, 𝑓𝑖 is the frequency of incident outcome case i, from the frequency analysis and
𝑝𝑓,𝑖 is the probability of that incident outcome case i that will result in a fatality at location x, y.

The societal risk of people affected by all incident outcome cases can be estimated using
the following equation (Eq. 24) [47].

𝑁𝑖 = �𝑥,𝑦 𝑃𝑥,𝑦 𝑝𝑓,𝑖 (24)

 Where, 𝑁𝑖 is the number of fatalities resulting from an incident outcome case i; 𝑃𝑥,𝑦 is the
number people at locations x, y; and 𝑝𝑓,𝑖 is the probability of that incident outcome case i will
result in a fatality at location x, y.

The risk assessment of the IOCL Jaipur accident has been carried out by using DNV
Norway based Phast Risk 6.7 software [23]. The study involves analysis of the impact of
overpressure due to vapour cloud explosions (VCEs) and the thermal radiation owing to tank
fires on the surrounding people and facilities. The VCEs have the potential to cause significant
knock-one effects. The effects of secondary events have also been included in the study. The
results of the risk modelling show the severity of incidence in terms of individual and societal
risk contours.

Individual Risk

The severe impact of the accident was expected due to the formation of large amounts of
air mixed flammable vapour cloud and subsequent fires on tanks. In the IOCL terminal, the
peripheral distance from the released gasoline tank to adjacent tanks of gasoline, kerosene and
diesel were 15m, 55m and 75m, respectively. The maximum mass burning rate of the gasoline in
most of the tank fires had been about 0.055 kg/m2s [38].

Individual fatality risk levels reflect the cumulative risk implication of various events of
varying consequences and likelihood of occurrence. The tolerable or acceptable value of the
individual risk for personnel or industrial installations is not yet regulated by Indian standards
and norms. Therefore, a comparison of the calculated risk values was made with the tolerable /
acceptable risk values proposed by HSE UK guidelines. Maximum tolerable individual risk to
site workers as per HSE UK guidelines is 10-3 per year whereas the same for the public is 10-4
per year [48].

The individual risk has been computed by the DNV Norway based Phast Risk- 6.7
software for the territory of the terminal and the surrounding area. The individual risk contours
with various risk levels have been presented in Fig. 14. The maximum risk level of 10-4 per year
has been observed near the storage tank area at a distance of around 100m from the release point.
The next risk level i.e. 10-5 per year is at a distance of 280 m. These risk contours fall within the
terminal boundary. In this case, risk at the terminal does not lie in completely unacceptable
region as the level is not exceeding the value of 10-4per year.

As the risk levels of 10-4 and 10-5 per year corresponds to the ALARP region, the risk in
the terminal should have been minimized with more precautionary measures. The individual risk
outside the terminal is more than 10-6 per year making it as an acceptable risk level [50].

The risk level for the surrounding people was tolerable. Thus, the quantitative risk
assessment demonstrates clearly that the safety precautionary measures were not effectively
implemented in the terminal, which subsequently led to the severity of the accident. The results
indicate that the incident could have been avoided / minimized by the proper implementation of
safety measures. However, it seems that the failure of or absence of adequate precautionary
measures led to such a catastrophic accident.

[Source: 49]
Fig. 14 Individual Risk controls for the IOCL Jaipur incident
Societal Risk

The societal risk is presented as an F-N curve which is a plotting of cumulative frequency
versus number of fatalities. The X-axis indicates the number of fatalities and the Y-axis gives the
cumulative frequency (per year) of all the scenarios together. Fig. 15 shows an F/N curve for the
incident delineating three regions viz. “Unacceptable”, “tolerable if ALARP” and “broadly
acceptable”. Since the number of deaths and frequency cover several orders of magnitude, an
arithmetic plotting is generally used for this purpose. To evaluate the societal risk, which reflects
the acceptable individual risk criteria, it is significant to consider what the size of the population
is, over which the risk must be shared.

Due to unavailability of India specific values, as a reference to determine criteria for

socially acceptable safety level, the criteria used in foreign countries have been surveyed.
According to HSE UK guidelines, acceptable frequency level is less than 1×10−4 per year, the
buffer zone level lies in between 1×10−4 - 1×10−2 per year while unacceptable frequency level is
higher than 1×10−2 per year [46]. The F/N curve for the IOCL Jaipur accident is in the ALARP
region [50]. This region indicates that the risk to the surrounding population is tolerable if the
precautionary measures are properly implemented. Failure in the periodical maintenance of the
valves and properly implemented precautionary measures might have been the reason for this
accident.
 Fig. 15 F/N curve for Jaipur IOCL incident

Emergency Response Plan For Storage Terminals

In the petroleum storage industry, emergency response planners have focused on

designing better and safer equipment and self-contained self-rescuers, minimizing response time,
increasing training of rescue teams, and emerging escape plans that comply with storage site
safety regulations. Immediate and effective response to an accident site is, of course, necessary
to reduce the severity of accidents, loss of life and the possibility of the loss of the future
productivity of the storage site [15]. Thus, the main focus in the management of emergencies has
been on resources and logistics; in other words, having who and what you need, when and where
you need it to encounter the crisis within an urgent time frame. EMERGENCY RESPONSE
PLAN (ERP) is required for various types of accidents to decrease the degree of hazards
efficiently by effectively preparing, responding, and restoring normal conditions [15]. Designing
improved equipment along with the application of new technologies and focused training
increases the efficiency of rescue operations. Additionally, modified technology has brought
more efficient communication, such that personnel in the command centre have the opportunity
to apprehend real time scenarios as precisely as the front line emergency workers. An effective
communication system almost always curbs the severity of an emergency.

This study shows that information obtained from post risk assessment activities carried
out on the IOCL Jaipur accident has generated significant data that is essential in emergency
response planning. The data generated from the IOCL incident, considered to be crucial in
framing on-site emergency plan of storage terminals, is also necessary for an off-site plan.
Therefore, it can be said that a complete ERP must be effectively developed and distributed on
the basis of the real scenario to prevent major incidents in the future. Predictive techniques
enable major accident consequences to be assessed and thus aid in the development and
implementation of mitigatory strategies incorporated in an ERP.

Conclusions
The study carried out in this paper has focused on post-risk assessment of petroleum
storage site with an aim to utilize the findings for deploying appropriate preventive measures and
delineating Emergency Response Plan (ERP) so as to reduce and mitigate hazards posed by such
facilities. For this purpose, modeling and simulation of vapour cloud explosion (VCE), large tank
fire, and individual and societal risk have been carried out and applied for IOCL, Jaipur accident.
Vapour cloud explosions are highly complex phenomena whose destructive potential
depends on not only the flammable mass involved but also the cloud dispersion and the reactivity
of the gaseous mixture. Among those parameters, the concentration, size, and location of the
vapour cloud play important roles, which is evident from IOCL Jaipur (India) accident assessed
in this research work. The evidences obtained from the IOCL Jaipur site are consistent with the
observations made during the follow-up study after the Buncefield incident both in terms of
overpressure damage and directional indicators. The observed damage at the site can be
explained in terms of high-speed deflagrations and transition to detonations. Overpressures in
excess of 200 kPa (2 bar) were generated across the site within the terminal, which, however,
was not uniformly distributed throughout the terminal.

However, the severity of explosion has successfully been explained using the current
knowledge of vapour cloud explosions and information available in the open literature. The
overpressure damage and the directional indicators show that the flammable vapour cloud
covered almost the entire site. The widespread high overpressures and the directional indicators
in the open areas infer that the vapour cloud explosion might have not been caused by
deflagration alone. The overpressure damage and directional indicators show that the source of
the detonation most likely was in the Pipeline Division area in the northeast corner of the site.
Flame entering into the pipeline division area might have caused a confined or partially confined
explosion, which possibly led to detonation as it vented from the building. The possibility of
detonation due to the line of trees along the north wall of the pipeline division has been ruled out
because it is not deep at lower levels, and there were some gaps in the tree line.

Flame characteristics like height and temperature have been computed to realize the
intensity of fire. The consequence modeling also needs to be carried out to analyze its effect on
each individual and surrounding population. Thus, individual and societal risk has been
quantified considering population in and around terminal. The individual and the societal risk
estimates show that the risk levels to which population is exposed in and around the terminal do
not exceed the tolerable limits proposed by the HSE UK standards and norms. The estimated risk
at the terminal was under ALARP region where substantial measures for a risk reduction were
needed. ALARP region indicates that the risk to the surrounding population is tolerable if the
precautionary measures are properly implemented. It is felt that India, too, requires stringent
guidelines to periodically assess the risks in such facilities to place appropriate safety measures
so that disasters of above-discussed nature could be avoided and/or minimized.

Emergency Response Plan (ERP) to tackle such type of big accidents is an integral and
essential part of a loss prevention strategy. The Emergency Operations Center (EOC) and
 Emergency Management Computation System should be an important components of ERP. The
Incident Command System (ICS) is a "function" oriented approach to an emergency response.
The success of system like e-ICS depends on relevance and accessibility of information and
response timeliness provided by the system whereas the acceptability is governed by behavioral
tendencies of the users in terms of perceived task support, group coordination and personal
biases.

Acknowledgments
We are thankful to the Indian Oil Corporation Ltd. for providing accident’s data. The support
of the National Environmental Engineering Research Institute (NEERI) is gratefully
acknowledged. This work is part of the Ph.D thesis being pursued by the first author at Centre
for Excellence in Disaster Mitigation and Management, I.I.T. Roorkee.

References

1. The Buncefield Incident 11 December 2005, The final report of the Major Incident Investigation
Board Volume 1 and 2, Health and Safety Executive, UK, 2011 ISBN
978 0 7176 6270 8.
http://www.buncefieldinvestigation.gov.uk/reports/index.htm
2. Caribbean Petroleum Refining Tank Explosion and Fire, Chemical Safety Board, USA,
2011. http://www.csb.gov/.
3. IOC Fire Accident Investigation Report, Oil Industry Safety Directorate, Ministry of Petrolium
and Natural Gas (MoPNG), Government of India, 2011.
http://oisd.nic.in/index.htm.
4. Drysdale, D., The Buncefield Oil Depot Explosions 11th December 2005, Combustion Institute
British Section: Spring Meeting, 2010.
5. Taveau, J., The Buncefield explosion: were the resulting overpressures really unforeseeable?
Process Safety Progress, 2012. 31 (1): p. 55–71.
6. Taveau, J., Explosion of fixed roof atmospheric storage tanks, Part 1: Background and review of
case histories. Process Safety Progress, 2011. 30 (4): p. 381–392.
7. Gant, S. E., Atkinson, G.T., Dispersion of the vapour cloud in the Buncefield Incident. Journal of
Process Safety and Environmental Protection, 2011.89: p. 391–403.
8. Mishra, K.B., Wehrstedt, K. D., Krebs, H., 2013. Lessons learned from recent fuel storage fires.
Journal of Fuel Processing Technology, 2013.107: p. 166–172.
9. Sharma, R.K., Gurjar, B.R., Wate, S.R., Ghuge, S.P. Agrawal, R., 2013. Assessment of an
accidental vapour cloud explosion: Lessons from the Indian Oil Corporation Ltd. accident at
Jaipur, India. Journal of Loss Prevention in the Process Industries, 2013. 26: p. 82-90.
10. Yang, X., Dinh, L.T.T., D. Castellanos, Amado, C.H.O., Dedy, Ng., Mannan, M. S., Common
lessons learned from an analysis of multiple case histories. Process Safety Progress, 2011. 30 (2):
p. 143–147.
11. Johnson, D.M., The potential for vapor cloud explosions — lessons from the Buncefield. Journal
of Loss Prevention in the Process Industries, 2010. 23 (6): p. 921–927.
12. Herbert, I., The UK Buncefield incident — the view from a UK risk assessment engineer.
Journal of Loss Prevention in the Process Industries, 2010. 6: p. 913–920.
13. Kletz,T., What went wrong? Fifth edition Butterworth-Heinemann978-1-85617-531-9, 2001.
14. Mannan, S., Lees Loss Prevention in the Process Industries, Hazard Identification, Assessment
and Control, Third edition Butterworth-Heinemann, 2005. p. 7506-7858.
15. Kowalski, K. M., A human component to consider in your emergency management plans: the
critical incident stress factor. Journal of Safety Science. 1995, 20: p. 115–123.
16. Raj., P.K., A review of the criteria for people exposure to radiant heat flux from fires. Journal of
Hazardous Materials, 2008. 159: p. 61–71.
17. Mudan., K.S., Thermal radiation hazards from hydrocarbon pool fires. Progress in Energy and
Combustion Science, 1984. 10: p. 59–80.
18. Crowl, D.A., Louvar, J.F., Chemical Process Safety. Fundamentals with Applications. New
Jersey: Prentice Hall, Englewood Cliffs.1990.
19. CCPS (Center for Chemical Process Safety) of the American Institute of Chemical Engineers.
Guidelines for chemical process quantitative risk analysis, 2nd ed., Wiley inter science, New
York. 2000.
20. Matthiesen, J. C., Estimating Chemical Exposure Level in the Workplace. Chem. Eng. Prog,
1986. .p.-30.
21. Herbert, I., The UK Buncefield incident - The view from a UK risk assessment engineer. Journal
of Loss Prevention in the Process Industries. 2010. 23: p. 913-920.
22. Bradley, D., Chamberlain, G. A., Drysdale, D., 2012. Large vapour cloud explosions, with
particular reference to that at Buncefield, Phil. Trans. R. Soc. A. 2012. 370: p. 544–566.
23. DNV (Det Norske Veritas). (2011). “Phast release note for version
6.7.” http://www.dnv.com.cn/binaries/release%20letter%20phast_tcm142 -466407.pdf (Mar. 4,
2015).
24. Baker, Q.A., Tang., E.A., Scheier, E.A., Silva, G.J., Vapor cloud explosion analysis. Journal of
Process Safety Progress. 1996. 15 (2): p. 106–109.
25. http://www2.gexcon.com/calculators/new/TNTMethod.php.
26. V. Raghunathan: Recent Advances in Vapour cloud modelling for Offshore Installations, DNV
Energy, 2006.
http://www.dnv.nl/onzediensten/software/publications/ publishedpapers/index.asp.
27. Quest Quarterly Newsletter, Oklahoma 4 (1) (Spring, 1999) 1–4.
www.questconsult.com/99-spring.pdf.
28. Mishra., K.B., Experimental investigation and CFD simulation of organic peroxide pool fires
(TBPB and TBPEH), BAM-Dissertation Series 63, Berlin, Germany, 2010, ISBN 978-3-
9813550-6-2.
http://www.bam.de/en/service/publikationen/publikationen-medien/dissertationen/diss-63e-
vt.pdf.
29. Van den Berg., A.C., The multi-energy method: a framework for vapor cloud explosion blast
prediction. Journal of Hazardous Materials, 1989. 12: p. 1–10.
30. Hailwood, M., Gawlowski, M., Schalau, B., Schönbucher, A., 2009. Conclusions Drawn from
the Buncefield and Naples Incidents Regarding the Utilization of Consequence Models, Journal
of Chemical Engineering Technology, 2009. 32: p. 207–231.
31. Engelhard, W.F.J.M., Heat Flux from Fires, in : C.J.H. van den Bosch, R.A.P.M. Weterings
(Eds), Methods for the calculation of Physical Effects, Publicatiereeks Gevaarlijke Stoffen, The
Netherlands. 2005. 6.1- 6.130.
32. Lees, F.P., Loss prevention in the process industries, Vols. 1-2. London: Butterworths. 1980.
33. Johnson, D. M., Characteristics of the Vapour Cloud Explosion Incident at the IOC Terminal in
Jaipur, 29th October 2009, GL Noble Denton, Report Number. 2011.11510, 1-
17 http://new.fabig.com/Files/GL_Report_No._11510.pdf .
34. Bradley, D., Lawes, M., and Kexin, L., 2008. Turbulent flame speeds in ducts and the
deflagration/detonation transition. Journal of Combust and Flame, 2008. 154: p. 96–108.
35. Vela, I., CFD prediction of thermal radiation of large, sooty, hydrocarbon pool fires, PhD Thesis,
University of Duisburg-Essen, Germany, 2009.
http://duepublico.uniduisburg-essen.de/servlets/DerivateServlet/Derivate-22199/Thesis-Vela.pdf.
36. Rew, P. J., Hulbert, W. G., Deaves, D. M., Modeling of thermal radiation from external
hydrocarbon pool fires. Journal of Process Safety and Environmental Protection, 1997. 75: p. 81-
89.
37. Munoz, M., Planas, E., Ferrero, F., Casal, J., 2007. Predicting the emissive power of
hydrocarbon pool fires, Journal of Hazardous Materials. 2007, 144: p. 725–729.
38. Muñoz, M., Arnaldos, J., Casal, J., Planas, E., 2004. Analysis of the geometric and radiative
characteristics of hydrocarbon pool fires, Journal of Combustion and Flame. 2004. 139: p. 263–
277.
39. Planas, C. E., Casal, J., Modeling Temperature Evolution in Equipment Engulfed in a Pool-fire,
Journal of Fire Safety. 1998. 30: p. 251-268.
40. Koseki, H., Combustion Properties of Large Liquid Pool Fires. Journal of Fire Technology,
1989. 25: p. 241-255.
41. Croce, A. P., Mudan, K. S., Calculation impacts for large open hydrocarbons fires. Journal of
Fire safety, 1986. 11: p. 99-112.
42. Chuna, H., Wehrstedtb, K. D., Velac, I., Schönbucher, A., Thermal radiation of di-tert-butyl
peroxide pool fires—Experimental investigation and CFD simulation, Journal of Hazardous
Materials. 2009. 167: p. 105–113.
43. Babrauskas, V., Estimating Large Pool Fire Burning Rates, Journal of Fire Technology.1983. 19:
p. 251-261.
44. Jo, Y. D., Crowl, D. A., Individual risk analysis of high-pressure natural gas pipelines. Journal of
Loss Prevention in the Process Industries, 2008. 21: p. 589–595.
45. Institute of Chemical Engineering (Ichem), Nomenclature for hazard and risk assessment in the
process industries. 1985.
46. Health and Safety Executive (HSE), Reducing Risks, Protecting People-HSE’s Decision Making
Process, 2001. p. 1-74. (URL: http://www.hse.gov.uk/risk/theory/r2p2.pdf) (Last accessed:
22.2.13).
47. Renjith, V.R., Madhu, G., Individual and societal risk analysis and mapping of human
vulnerability to chemical accidents in the vicinity of an industrial area, Journal of Applied
Engineering Research, Dindigul, 2010. 1: p. 35-148.
48. Kauer, R., Fabbri, L., Giribone, R., Heerings, J., Risk Acceptance Criteria and Regulatory
Aspects, 2002. OMMI (Vol.1 Issue 2): p. 1- 11.
49. Google Earth, Google Earth (Version 6), 2011 Available
at:(URL: http://www.google.com/earth/download/ge/ ) (Last accessed: 30.6.2011).
50. Gurjar, B.R., Sharma, R. K., Ghuge, S. P., Wate, S.R., Agrawal, R., Individual and Societal Risk
Assessment for a Petroleum Oil Storage Terminal. Journal of Hazardous, Toxic, and Radioactive
Waste, American Society of Civil Engineers (ASCE).
http://dx.doi.org/10.1061/(ASCE)HZ.2153-5515.0000277